Study On The Application Of Ensemble Prediction In The Extended Range Forecast And Hybrid Data Assimilation

In order to promote the application of ensemble prediction in extended range forecast and data assimilation, two studies about the interpretation method of extended range(11-15day) forecast product based on T213 global ensemble prediction system and EnKF-3DV hybrid data assimilation system for GRAPES_Meso model are carried out. Thus to provide some reference for the development of the national numerical weather prediction system.Based on the 11-15 day forecast data of T213-GEPS 500 hPa geopotential height and 850 hPa temperature in China and the adjacent region at January 2008, an interpretation method of extended range forecast product is designed, the pentad average and pentad anomaly forecast of extended range(11-15day) are got after the 11-15 day daily forecast data are processed by ensemble mean、moving average of backward decaying weighting coefficient after terminal moving average and pentad average. Meanwhile, the forecast effect of the pentad average and pentad anomaly forecast is verified. Results show that :Taken as a whole, the forecast effect of 11-15 day pentad average is good enough to be used. Meanwhile, the 11-15 day pentad anomaly forecast has strong skill in the forecast of persistent large-scale anomalous anomaly. The position, distribution and center of anomalous anomaly can all be forecasted well but the overall intensity is weaker than observed field.Based on the GRAPES regional ensemble prediction system and the 3Dvar data assimilation system that are implemented operationally at the numerical weather prediction center of Chinese meteorological administration, an EnKF-3DV hybrid data assimilation system for GRAPES_Meso model is established using the extended control variable technique to implement a hybrid background error covariance that combines the climatological covariance and ensemble-estimated covariance. Aiming at the problems of EnKF data assimilation part of the system include the degree of geostrophic balance between variables is reduced, the analysis increment is not smooth and the analysis increment is obviously smaller than the 3Dvar data assimilation, corresponding measures are taken to optimize and ameliorate the system. Based on this, a single pressure observation EnKF data assimilation experiment is conducted to ensure the EnKF data assimilation part of the system is correct and reasonable, some localization scale sensitive tests of EnKF data assimilation are conducted to determine the most appropriate localization scale. On that basis, some hybrid data assimilation experiments are conducted. Results show that: it is the most appropriate to set the weight factors of ensemble-estimated covariance to 0.8 in the experiment. Contrasting to the 3Dvar data assimilation, the geopotential height forecast of the hybrid data assimilation experiments improve very little, but the winds forecast has a slight improvement at each forecast time especially over 300 hPa. Overall, the hybrid data assimilation has some advantage over 3DVar data assimilation.